Integer wavelet conversion method based on digital signal processor

Abstract

The integer wavelet transformation method based on digital signal processor is characterized by that when the wavelent is undergone the process of positive transformation, firstly, dividing the image data into several data sections, then respectively making line data of current data section undergo the process of line positive transformation, and making column data of current data section undergo the process of column positive transformation, and when the vavelet is undergone the process of inverse transformation, firstly, dividing the image data into several data sections, then respectively making column inverse transformation of column data of current data section and line inverse transformation of line data of current data section, crossly making line and column transformation until the whole image data are processed completely.

Description

technical field [0001] The invention relates to a wavelet transform method, in particular to an integer wavelet transform method based on a digital signal processor. Background technique [0002] At present, wavelet transform has been used as the international standard of JPEG2000, and is widely used in static texture information compression and medical image compression of JPEG2000 and MPEG4. The multi-resolution analysis of wavelet transform, the ability to truncate the embedded code stream, and no mosaic have advantages that traditional discrete cosine transform (DCT) cannot match. However, wavelet transform operates on the entire image, not like DCT transform. The operation is divided into blocks, so the parallelism of the operation in the transformation is not high, and it is very difficult to implement hardware. [0003] However, in many practical systems, such as digital cameras, videophones, camcorders, and handheld computers, considering the requirements of speed a...

AI Technical Summary

Problems solved by technology

[0007] 2. The operation is performed on the entire image, and the memory space of the hardware DSP is very limited. When the image size is large, the image data can only be placed in the off-chip storage space, which greatly reduces the operation speed

Taking the CIF format as an example, the size of an image is 352×288×2=198KB, and the on-chip RAM of TI C6000 DSP cannot meet the needs, and when the image size is larger, the lack of memory space is even worse. protrude

[0008] 3. In the wavelet transform of a two-dimensional image, a coefficient needs to be called repeatedly at least twice (row filter once, column filter once), frequent on-chip and off-chip data scheduling reduces the execution efficiency of the algorithm

[0009] 4. The quantization operation is performed separately after the wavelet transform is completed, requiring M×N loop operations, which greatly increases the execution time of the operation

[0010] 5. Many wavelet transform filter parameters are floating-point numbers, and the operation of floating-point numbers is very time-consuming for hardware, which affects the realization of real-time performance

[0011] Although some improvements have been made in response to the above defects, for example, the floating-point operation of the wavelet transform is changed to the integer lifting operation, and the multiplication and division operation is replaced by a shift and addition operation; However, these improvements only improved the operational efficiency of wavelet transform from the perspective of calculation formula, and did not fundamentally solve the problem of low parallelism of overall operation and memory consumption encountered by wavelet transform in hardware implementation. , data repetition scheduling and other issues

Images